Urban landscape genetics of white-footed mice

Since 2008, my laboratory has been developing the white-footed mouse (Peromyscus leucopus) as a model system to study the evolutionary impacts of urbanization in the NYC metro area.  We focus on this species because it occurs in nearly all forested patches in NYC, is easy to trap, and its population ecology is well-known.  The genomes of multiple Peromyscus species are currently being sequenced.  Our research employs landscape genetics, population genomics, and long-term field studies to examine the modes and tempo of evolution in New York City mice

urban landscape genetics

Our study sites in NYC (red dots). Shades of green represent percent canopy cover.

Despite intense urbanization, more than 20% of NYC is still comprised of green spaces.  These remnant forests and other types of semi-natural vegetation (largely city parks) are unevenly distributed and highly fragmented, but do contain a mixture of native wildlife and introduced species that together constitute a novel ecosystem.  City parks also represent a “natural” experiment of the genetics of metapopulations in human landscapes.  The emerging field of landscape genetics combines landscape ecology and population genetics to build models that account for habitat permeability, barriers, and other aspects of the physical environment that influence the movement of animals and their genes.  In 2008, I was awarded a National Science Foundation grant to conduct the first-ever landscape genetic analyses of wildlife inhabiting the urban core.

From 2008-09, we trapped over 300 mice at 15 sites and genotyped them at 18 microsatellite loci to examine neutral evolution in urban Peromyscus.  Much to our surprise, both classic indices and Bayesian clustering analyses indicated strong population structure throughout the city, and identified most city parks as harboring unique evolutionary populations (Munshi-South & Kharchenko, 2010).  However, heterozygosity and effective population size remain high.  Peromyscus spp. typically exhibit minimal genetic structure over even broad geographic scales, indicating that urbanization in NYC is a potent driver of genetic differentiation.

STRUCTURE bar plot from NYC white-footed mice (Munshi-South & Kharchenko, 2010). Each bar represents the genetic data from one individual mouse, and the colors correspond to inferred evolutionary clusters. Most mice in NYC were assigned to unique evolutionary clusters correspond to the park in which they were trapped. Numbers on top are sample sizes.

Circuitscape model of connectivity among Bronx and Manhattan populations of white-footed mice. Resistance values were based on percent canopy cover seen in the study site map above. Lighter areas indicate greater connectivity of the landscape. These models explained nearly 90% of the variation in recent gene flow between these populations (Munshi-South, 2012).

To examine the influence of the urban landscape in creating this genetic structure, we first estimated both recent and historical migration rates between urban white-footed mouse populations.  Next, we created two types of landscape connectivity models: least-cost paths that identify the single best path through the landscape (i.e. isolation-by-effective-distance), and isolation-by-resistance models that identify all potential paths using a new implementation of circuit theory for landscape ecology (i.e. Circuitscape).  After exploring several landscape variables, we used percent tree canopy cover as the best predictor of connectivity between white-footed mouse populations.  Both least-cost path and isolation-by-resistance models were significantly associated with migration rates, even after factoring out the influence of isolation-by-distance.  However, in all cases isolation-by-resistance models outperformed the others, and explained nearly 90% of the variation in recent migration rates (Munshi-South, 2012).  While often mentioned in landscape genetics review papers, the isolation-by-resistance approach has rarely been implemented because of its high computational demands.  These models also identified specific corridors of gene flow through the city, such as cemeteries, roadway medians, and other relatively green infrastructure that are of interest to park managers and wildlife biologists.

Mismatch distribution of mtDNA haplotypes showing population expansion after a recent demographic event (or selection).

Having identified pervasive genetic structure and the landscape characteristics that promote connectivity among white-footed mice in an urban landscape, we are now investigating the historical demography of these populations.  An undergraduate independent study in my lab used mismatch distribution analysis of mitochondrial haplotypes to show that these populations experienced a major demographic event (or selection) in the last few hundreds of generations.  Similarly, preliminary Approximate Bayesian Computation (ABC) approaches support models of rapid divergence between urban populations on the timescale of urbanization in NYC.  Our ongoing ABC modeling and analysis of park size vs. genetic diversity will be included in a forthcoming manuscript on historical demography in NYC.